21 Comparison of the NIH Toolbox Cognition Battery to Established Performance-Based Assessments in a Pediatric Cancer Setting

Abstract

Objective:

This study examines the clinical validity of the NIH Toolbox Cognition Battery measures in patients with oncological diagnoses and tumor predisposition syndromes, including Neurofibromatosis, Type 1 (NF1).

Participants and Methods:

Participants included 158 patients (61% male, 67% White) ages 3 to 25 years (M = 8.38, SD = 4.32) who underwent neuropsychological evaluation between 2019 and 2022. Patients with brain tumors (n = 50) and leukemias (n = 49) accounted for 2/3 of the sample. The remainder had solid tumors, lymphomas, or cancer predisposition syndrome. Forty-eight had a diagnosis of NF1. Performance-based measures of attention, executive functioning, and processing speed were administered as part of neuropsychological evaluations. Patients were administered between 1 to 4 NIH Toolbox Cognition measures, including Flanker Inhibitory Control and Attention Test (Flanker), Dimensional Change Card Sort Test (DCCS), Pattern Comparison Processing Speed Test (PCCS), and List Sorting Test. Parent-reported measures of attention and EF were also obtained. Z-scores were used to compare performance across measures that assessed equivalent constructs. The rates of weak performance (>1 SD below the mean) using Toolbox measures were compared to rates of weak performance on traditional neuropsychological measures (e.g., Digit Span), and rates of functional impairment (e.g., parent-reported concerns, ADHD diagnosis).

Results:

FSIQ, Coding, and NEPSY Inhibition correlated with all 4 Toolbox measures, while Digit Span correlated with List Sorting, DCCS, and Flanker. DCCS and PCCS correlated with verbal fluency measures. NF1 patients scored lower than non-NF1 patients on Flanker, F(1,126) = 13.01, p<.001 and DCCS, F(1,150) = 6.85, p = .01. Toolbox performance did not differ significantly by age group. Rates of identified weakness were relatively similar on Toolbox measures, some traditional measures, and parent-reported attention problems. In identifying those with and without weakness, the agreement between Flanker and other measures ranged from 52% (Auditory Attention) to 66% (Coding). Agreement between DCCS and traditional measures ranged from 47% (Letter Fluency) to 80% (Switching). For PCCS, concordance ranged from 45% (Semantic Fluency) to 69% (Switching). List Sorting had 80% agreement with Digit Span and Coding.

List Sorting had the highest agreement with parent-reported attention problems (76%), EF problems (72%), and ADHD diagnosis (79%). There was relatively high concurrence between DCCS and ADHD diagnosis (69%) and parent-reported attention problems (60%) and EF problems (65%) and between Flanker and ADHD diagnosis (67%). PCCS had less agreement with functional outcomes, ranging from 49% for EF problems to 58% for attention problems and ADHD diagnosis. In comparison, Digit Span had 64% agreement for EF problems, 70% for attention problems, and 73% for ADHD diagnosis.

Conclusions:

The NIH Toolbox Cognition Battery can be used to identify neurocognitive weaknesses in pediatric oncology patients and provide clinically meaningful data. Evaluation of the Toolbox measures’ sensitivity to change over time is warranted, as monitoring the progression of cognitive late effects is particularly salient in cancer survivorship.